Rudzani Muloiwa
MBChB, (Natal), DCH (SA), FCPaed (SA), MSc (LSHTM)
Thesis Presented for the Degree of DOCTOR OF PHILOSOPHY in the Department of Paediatrics and Child Health,
Faculty of Health Sciences, UNIVERSITY OF CAPE TOWN
February 2020
Supervisors
Prof Heather J. Zar, University of Cape Town Prof Gregory D. Hussey, University of Cape Town
Epidemiology of pertussis in children
hospitalised with respiratory tract infection
University
of Cape
Town
The copyright of this thesis vests in the author. No quotation from it or information derived from it is to be published without full acknowledgement of the source.
The thesis is to be used for private study or non- commercial research purposes only.
Published by the University of Cape Town (UCT) in terms of the non-exclusive license granted to UCT by the author.
University
of Cape
Town
For he was a shrub among the poplars Needing more roots
More sap to grow to sunlight, Thirsting for sunlight,
A low growth among the forest.
Into the soul
The selves extended their branches, Into the moments of each living hour, Feeling for audience
Straining thin among the echoes;
And out for the solitude
Voice and soul with selves unite, Riding the echoes,
Horsemen of the apocalypse;
And crowned with one self The name displays its foliage, Hanging low
A green cloud above the forest.
Christopher Ifekandu Okigbo (Siren Limits II)
Declaration
I, Rudzani Muloiwa, hereby declare that this thesis is my own work, both in concept and execution, apart from the normal guidance received from my supervisors and contributions from others as outlined in the Introduction and the acknowledgement section of each chapter. The assistance I received with study management, data collection, analysis and manuscript review from the co-authors of the publications that form part of this thesis is described for each relevant chapter.
Neither the substance nor any part of the above thesis has been submitted in the past, or is being, or is to be submitted for a degree at this University or at any other university.
I grant the University of Cape Town free license to reproduce the above thesis in whole or in part, for the purpose of research.
I present this thesis for examination for the degree of PhD.
Signed: Dated: 29th December 2020
Contents
List of Tables ... 6
List of Figures ...7
Abstract ... 9
Style and abbreviations ... 13
Introduction ... 16
The burden of laboratory-confirmed pertussis in low- and middle-income countries since the inception of the Expanded Programme on Immunisation (EPI) in 1974: a systematic review and meta-analysis ... 36
Incidence and diagnosis of pertussis in South African children hospitalised with lower respiratory tract infection ... 74
Risk factors for childhood Bordetella pertussis disease in hospitalised children...93
Co-detection of Bordetella pertussis and other respiratory organisms in children hospitalised with lower respiratory tract infection ... 113
Diagnostic limitations of clinical case definitions of pertussis in infants and children with severe lower respiratory tract infection ... 134
Highlights and Conclusions ... 153
Appendix A: Informed consent form ... 156
Appendix B: Ethical approval HREC 371/2011 ... 160
Appendix C: Case Report Form ... 162
Acknowledgments ... 11
List of Tables
Chapter 1
Table 1: Comparison of diagnostic methods for Bordetella pertussis ……….………..……..…...21
Chapter 2
Table 1: Strategy used to search for literature in MEDLINE (Via Pubmed)…..………..…38 Table 2: Characteristics of studies included in the systematic review……….…...43 Table 3: Population and hospitalisation incidence rates of Bordetella pertussis..……….50
Chapter 3
Table 1: Baseline characteristics of the study participants, Bordetella pertussis cases
and age matched NP controls…………...….……….………....79 Table 2: Bordetella incidence stratified by age group, HIV and vaccine doses………...82
Chapter 4
Table 1:Baseline characteristics of enrolled children………….………..….98 Table 2: Caregiver characteristics by child’s B. pertussis PCR status………...…...100 Table 3: Risk factors for confirmed Bordetella pertussis infection in study children………....102
Chapter 5
Table 1: Baseline characteristics of study participants……….….….……..…..119 Table 2: Association between Bordetella pertussis and other organisms isolated on IS……….…...122 Table 3. Risk of lower respiratory co-infection in children with confirmed
Bordetella pertussis infection ……….………..……..123
Chapter 6
Table 1: Clinical features for diagnosis of pertussis cases……….…...136 Table 2: Baseline characteristics of study participants………..……….140 Table 3: Clinical presentation of children by Bordetella pertussis PCR status….……..……...……141
List of Figures
Chapter 2
Figure 1. Studies included in the systematic review………..………..………..…42 Figure 2. Prevalence of polymerase chain reaction confirmed Bordetella pertussis………...46 Figure 3. Prevalence of culture confirmed Bordetella pertussis………..………..47 Figure 4. Distribution of point prevalence of polymerase chain reaction and culture confirmed
pertussis by period (a) and age group (b)………..……….49 Figure 5. Meta-analysis of the relative risk of pertussis comparing HIV unexposed
uninfected (HUU) to HIV exposed uninfected (HEU) (a) and HIV infected (b)...52 Figure 6. Mortality and case fatality rate of confirmed pertussis………...53 Additional file 1: Country and year of included studies with confirmed pertussis shown by
World Health Organization region……….………...67 Additional file 2: Distribution of point prevalence of confirmed pertussis by World Health
Organisation region and confirmation……….………..67 Additional file 3: Prevalence of paired serology confirmed Bordetella pertussis………..……...68 Additional file 4: Prevalence of polymerase chain reaction and culture confirmed Bordetella parapertussis.…69 Additional file 5: Meta-analysis of relative detection rates of polymerase chain reaction
and culture in confirming Bordetella pertussis infection……….70
Chapter 3
Figure 1. Recruited lower respiratory tract infection cases showing number and Percentage of
confirmed Bordetella per month………..………..….81
Chapter 4
Figure 1. Enrolment flow diagram of study participants showing number of Bordetella pertussis
positive children and caregivers………...……….97
Chapter 5
Figure 1. Distribution of number of bacteria (a) and bacteria + viruses (b) identified on polymerase chain reaction (PCR) in participants with and without Bordetella pertussis………..………...121 Figure 2. Proportion of children with hypoxaemia and chest indrawing by presence of co-detected
organisms that were strongly associated with pertussis………125 Figure 3. Length of hospital stay by presence of co-detected organisms strongly associated with pertussis..…126
Chapter 6
Figure 1. Sensitivity and specificity of clinical features in the diagnosis of pertussis……….142 Figure 2. Receiver operating characteristics (ROC) curves for duration of symptoms………...143 Figure 3. Sensitivity and specificity of clinical features in the diagnosis of pertussis with changing duration...144 Figure 4. Sensitivity and specificity of lymphocytosis in the diagnosis of pertussis……….………..145
Abstract
The availability of an effective vaccine against Bordetella pertussis substantially reduced the morbidity and mortality from pertussis, however, in the last decade there appears to have been a substantial increase in pertussis cases as reported mainly in high income countries.
Although it is believed that the greatest burden of pertussis, including deaths, is in low- and middle-income countries (LMICs), there seem to be little data available to back this up.
This thesis set out to find data that will give some insight into the burden of pertussis in a low- and middle-income setting in infants and children with severe lower respiratory tract infection (LRTI). Given the paucity of data in LMICs, the thesis started by systematically searching for existing data that will give some indication of the possible extent of the pertussis problem in these countries. Secondly, a prospective study was conducted at a children’s hospital. As hospital admission is a marker of severe disease, these children were targeted as the appropriate population in which to meaningfully conduct a primary study on the burden of pertussis. In addition to quantifying the burden by describing the prevalence of confirmed pertussis in this group of children, the study set out to look for potential factors that may be associated with increased risk of pertussis. LRTI are now commonly known to be associated with identification of multiple organisms in respiratory samples, this study aimed to also look at organisms that are detected with Bordetella pertussis; and investigate whether this association was in any way associated with severe disease or negative outcomes.
Finally, this study hoped to identify clinical features that could be used to develop a more reliable clinical case definition of pertussis.
Chapter 1 gives a background that justifies the undertaking of this study. In chapter 2 a
systematic review quantifies, using the best available data, the burden of pertussis in LMICs.
Chapter 3 clarifies the methods briefly described in the rest of the manuscript. The burden of pertussis due to the two organisms known to cause the disease, Bordetella pertussis and Bordetella parapertussis, is described in some detail. In both this chapter and the earlier
mentioned systematic review (chapter 2), the burden of pertussis is stratified by subgroups to identify potential risk factors. The issue of risk is formally and specifically taken up in the chapter that follows (chapter 5) where potential risk factors are analysed, and the independent impact for some of these factors is established.
The last two results chapters (chapters 6 and 7) deal respectively with the conundrum of finding other respiratory organism in the same specimen with Bordetella pertussis and failure to find useful clinical criteria that can help with improved diagnosis of pertussis.
While there is no established pattern noted between pertussis and most organisms, a few give signals of being independently associated with Bordetella pertussis even if the clinical relevance is not clear at the moment.
In the final chapter of the thesis (chapter 8) I conclude the thesis by making an argument that although there are still knowledge gaps, the thesis gives a clear indication that pertussis remains a serious problem in LMICs especially for some groups that show increased risk of the disease or its severe consequences.
Acknowledgements
Gratitude
To Heather and Greg, the night guides only believed at sunrise
To the duet that hummed incessantly and quietly, “This is not your name”
…and to Mark, Mark who believes that twice-watered cows will calf and give milk To Mugo and Sizwe and Emmanuel, who sang with me until they were hoarse
Gratitude
To the Physician Partnership Trust
Twice named in a single pulse for the unshattered long hearts in memorium
The pathfinders whose lifting is etched in stone (Maybe not all that is broken turns to dust) All remembering is indeed honey and gall
To Sanofi with my grandmother’s hands,
Her voice straining, not knowing how to cross the ocean:
“You do not live for me”
Gratitude
For the unnumbered gifts of blood and time Each freely given whilst holding breath
To Chris, Nomawethu and Nezisa, painstakingly counting each ounce of air
Gratitude
To Ngina and Vele for the neglect borne in silence
For the many moons that knew no other life but the distance
To the makers on their bruised knees, chanting my name from infinity to infinity For the gift of my callused hands: the courage to hold on to my soul
And here I am finally, at the morning-sunset Each guarded hair accounted for
The yeast is indeed hidden in the dough Gratitude
Style and abbreviations
The papers that have been included in this thesis comprise manuscripts submitted to both American and British journals. To maintain a sense of consistency through the thesis, all spellings in the manuscripts have been changed to comply with British English spellings, the spelling format most commonly followed in South Africa. The only exception to this has been in the included figures where the text within figures has been retained as submitted to the journals. Although all the journals to which manuscripts have been submitted use Vancouver referencing, they differ slightly in how the style is formatted. To maintain consistency throughout the thesis, a generic Vancouver style template has been adopted and applied throughout. In general, all the manuscripts that contribute to the various chapters have been reproduced in the thesis as submitted to the various journals. Each chapter therefore contains its own relevant literature review and acknowledgements.
Abbreviations:
aP - the acellular vaccine aRR - adjusted relative risks ART - antiretroviral treatment AUC - Area under the curve CDC - Centre for Disease Control 95% CIs - 95% confidence intervals DAG - directed acyclic graph
DPT - Diphtheria, Pertussis, Tetanus
EPI - National Expanded Program on Immunisation
GPI - Global Pertussis Initiative HEU - HIV-exposed uninfected HICs - High Income Countries HIV+ - HIV infected
HREC - Human Research Ethics Committee HUU - HIV-unexposed uninfected
ICH - Institute of Child Health IQR - interquartile ranges IS - induced sputum
LMIC - low- and middle-income countries LRTI - lower respiratory infection
MDI - metered dose inhaler MeSH - medical subject heading
NHLS - National Health Laboratory Services
NICD - National Institute of Communicable Disease NP - nasopharyngeal
PCR - polymerase chain reaction confirmed
RCH - Red Cross War Memorial Children's Hospital ROC - Receiver operating characteristics
RR - relative risks
RTHC - Road to Health Card UCT - University of Cape Town WAZ - weight for age Z scores WHO - World Health Organization wP - whole cell vaccine
Chapter 1
Introduction 1
2
Epidemiology of pertussis 3
According to the World Health Organization (WHO), there are more that 20 to 40 million 4
annual cases of pertussis, and 300,000 associated deaths due to the disease; 90% of which 5
are estimated to occur in low- and middle-income countries (LMIC) [1, 2]. Most 6
morbidity and mortality is seen amongst unimmunised or incompletely immunised 7
infants, who have more severe disease and are more likely to have complications.[3]
8
Increasingly, pertussis is also recognised as an important cause of disease in adolescents 9
and adults with waning immunity. Older individuals have less severe disease and fewer 10
complications, but substantial economic costs are associated with unrecognised infection 11
in these individuals who also serve as an important source of infection for non-immune 12
infants. [4]
13 14
Pertussis is a notifiable disease in South Africa (SA). Notification may be on clinical 15
suspicion alone and does not require laboratory confirmation, but laboratory tests should 16
be performed where available. At present there is no active surveillance for pertussis and 17
the introduction of such surveillance is not currently a priority. Lack of surveillance is not 18
a uniquely South African problem, most LMICs lack resources for surveillance.[4, 5]
19
Only 60 cases of pertussis were notified to the Department of Health from January 2000 to 20
September 2004, which is likely a substantial underestimate of the true prevalence of 21
disease in South Africa. Anecdotal reports from concerned paediatricians and general 22
practitioners indicate that the incidence of the disease is possibly much higher than the 23
statistics reflect.
24 25
A retrospective folder review of children at the Red Cross War Memorial Children’s 26
Hospital in Cape Town, South Africa revealed that 61 out of 75 (81%) children with 27
polymerase chain reaction confirmed (PCR) pertussis seen between 2008 and 2012 were 28
never notified. [6]These were children who were ill enough to warrant an investigation by 29
the attending physician. The epidemiology of pertussis in children presenting with less 30
severe respiratory disease who do not warrant admission is still unknown and further 31
study needs to be done in this regard.
32 33
Pathogenesis 34
Pertussis is an acute, communicable infection of the respiratory tract caused by Bordetella 35
pertussis and occasionally by Bordetella parapertussis. Both organisms are strict human 36
pathogens with most of the burden attributable to Bordetella pertussis. Other Bordetella 37
species have been known to cause human disease, but it is Bordetella holmesii that in 38
addition to Bordetella pertussis and Bordetella parapertussis has been recognised as a 39
cause of pertussis-like illness.[7] Pertussis is spread by droplets from person to person and 40
patients are infectious from 7 days after exposure to 3 weeks after the onset of 41
paroxysms.[8] The organism does not invade systemically but attach to ciliated epithelial 42
cells of the airways causing ciliastasis, local tissue damage and interference with 43
phagocytic cell functioning. Viscous secretions and sloughed cells may accumulate in the 44
airways and cause obstruction. Complications include apnoea, bronchopneumonia, otitis 45
media, atelectasis, pneumothorax, hypoxic seizures, encephalopathy, feeding difficulties, 46
and vomiting. Pressure-related complications include rectal prolapse, petechiae, hernias, 47
epistaxis, subconjunctival and intracranial haemorrhages.[4]
48 49
Central to the pathogenesis are the many toxins that are produced by the organism.[9] The 50
most important of these is the pertussis toxin that seems crucial in the pathogenesis of 51
severe disease and death.[10] Although Bordetella pertussis and Bordetella parapertussis 52
cause a clinically indistinguishable disease, the latter does not produce the pertussis 53
toxin.[7]
54 55
Diagnosis 56
Diagnosis of Bordetella pertussis illness in infants and young children is difficult as 57
clinical presentation is variable and non-specific. Diagnosis is made on the basis of the 58
clinical picture (mild cough, coryza and fever, progressing to paroxysmal cough, 59
precipitated by crying, eating or drinking, “whooping”, vomiting, cyanosis, sweating, 60
prostration and exhaustion).
61 62
In the absence of access to laboratory confirmation, most low and-middle income 63
countries rely exclusively on clinical criteria to diagnose pertussis. The two commonly 64
used diagnostic criteria are those defined by WHO and the Centers for Disease Control 65
and Prevention (CDC). Both WHO and CDC criteria include presence of a cough for at 66
least 14 days characterized by one of paroxysms, inspiratory whoop or post-tussive 67
vomiting.[11] In addition to the three clinical features CDC also includes presence of 68
apnoea in its criteria.[12] Disease presentation may be modified by age, previous 69
immunisation or infection, antibiotic exposure and concurrent infection with other 70
pathogens. As a result, the presentation of pertussis is frequently atypical, especially in 71
very young infants and adults. Supporting laboratory investigations are also important 72
(leucocytosis and lymphocytosis, isolation of B pertussis from nasopharyngeal 73
secretions).[13]
74 75
A conference poster presentation of a review of PCR 115 confirmed pertussis cases from 76
the area of Bloemfontein, South Africa, found a much shorter duration of cough (Median 77
6 days, interquartile range 3 -12days) that might not have met most clinical criteria for 78
suspicion of pertussis disease. Lobar pneumonia was a finding in 62% of the patients in 79
this cohort.[14]
80 81
Culture of the organism, which has been regarded as the gold standard, is possible but 82
difficult. Although culture is highly specific, it has poor sensitivity, particularly late in 83
disease. Culture is most likely to be positive during the catarrhal phase (5 to 14 days from 84
the onset of illness).[4] Most cases of pertussis are only recognised once they have a 85
paroxysmal cough. Unfortunately - especially when classical case definitions based on 86
long duration of cough symptoms are employed - the sensitivity of culture is very poor by 87
the time the diagnosis is suspected. Sensitivity is higher in infants than in adolescents and 88
adults and is influenced by quality and timing of specimen collection and laboratory 89
expertise.[4] Specimens should include a properly performed nasopharyngeal aspirate or 90
nasopharyngeal swab, ideally inoculated directly onto a suitable culture medium (e.g.:
91
Regan Lowe medium) at the bedside. The organism is fastidious and grows slowly.
92
Culture plates must be incubated for at least 7 days before reported negative. Culture 93
allows for surveillance of antibiotic resistance and molecular epidemiological typing in 94
outbreak situations.[4]
95 96
PCR has greatly improved the ability to confirm pertussis cases and is increasingly used 97
for diagnosis on clinical specimens. There are many advantages associated with the use of 98
PCR: results are rapid, less dependent on delays in transport and even non-viable 99
organisms may be detected. PCR targets include IS481, the common Bordetella target that 100
includes Bordetella pertussis and IS1001 for Bordetella parapertussis. The qualitative 101
PCR using these gene targets can be performed on a nasopharyngeal swab (Dacron not 102
calcium alginate swabs as the latter may inhibit PCR) and/or aspirate specimens. The 103
IS481 target is also found in other Bordetella species such as Bordetella holmesii and 104
Bordetella brochiseptica.[15] IS481 is therefore not specific to Bordetella pertussis. A 105
published study demonstrated that up to 20% of patients initially diagnosed as pertussis 106
using the IS481 target were in fact Bordetella holmesii infected[16]. If IS481, a target 107
with high sensitivity is used to screen cases, it remains important to test the positive 108
specimen further for the presence of pertussis toxin promoter gene sequences, as these are 109
specific for Bordetella pertussis.[15] Ideally two targets should be used for diagnostic 110
consensus. The sensitivity of PCR is highest earlier in the course of disease and declines 111
with time from the onset of symptoms. PCR is more sensitive than culture and remains 112
positive even once treatment is commenced. It is therefore still useful for persons 113
presenting as late as three weeks since onset of illness.[17]
114 115
In a prior unpublished study[18] to compare a standard nested PCR and a LightCycler- 116
based real-time assay with culture for confirming the diagnosis of pertussis, 48 children 117
presenting to Red Cross War Memorial Children’s Hospital with clinical features of 118
pertussis were sequentially enrolled, nasopharyngeal aspirates were collected, and 119
inoculated onto charcoal agar for isolation of Bordetella pertussis. The identity of colonies 120
morphologically resembling B. pertussis was confirmed by amplification of the pertussis 121
toxin promoter gene (ptxA-Pr). A nested PCR assay targeting the IS481 sequence was 122
performed directly on the NPA samples. In addition, a real-time PCR assay for detection 123
of the pertussis toxin promoter gene was performed on all samples. Bordetella pertussis 124
was cultured from five (10%) patients. However, the nested PCR assay for IS481 was 125
positive in 31 (65%) patients, and the toxin promoter gene was present in 17 (55%) of 126
these 31. No patient with a negative IS481 assay was culture positive or PCR positive for 127
the toxin promoter. These initial results suggest that PCR is a more sensitive method of 128
detecting Bordetella pertussis than culture. But it also shows that ptxA-Pr, even though 129
more specific for Bordetella pertussis, may lack the requisite sensitivity for case 130
confirmation. In the absence of double assays, PCR for the IS481 target, even though less 131
specific, may offer the best diagnostic confirmation support for clinically suspected cases 132
of pertussis.
133 134
Serological diagnosis using ELISA is available, quick and easy to perform in the 135
laboratory. As the measured response is uses antibodies against common vaccine 136
components, in patients recently vaccinated, it is not easy to distinguish between acute 137
infection and recent infection on a single serum sample.[19] Acute and convalescent 138
paired sera can be taken and the change in titres used for diagnostic purposes. Serology 139
suffers a number of drawbacks: only anti-PT (pertussis toxin) serology has been well- 140
standardized and validated, and only anti-PT IgG ELISA testing is recommended for use 141
in the diagnosis of pertussis. Appropriate cut-offs have not yet been determined in many 142
instances, making interpretation, especially for unpaired testing, difficult. IgG titres need 143
to be interpreted in the context of a diagnostic cut-off determined by local sero- 144
epidemiological surveys. A recommendation for considering anti-PT IgG titres between 145
50—120 international units per millilitre (IU/ml) as highly suggestive of recent pertussis 146
has been proposed by a collaboration of European laboratories, but this is based on sero- 147
epidemiological surveys in Western Europe.[20] No sero-epidemiological data exists for 148
SA or other countries in Sub-Saharan Africa from which we could infer appropriate cut- 149
off levels; this limits the usefulness of such tests in our setting at present.
150
Direct fluorescent antibody tests for direct antigen detection have poor sensitivity and 151
specificity and should not be relied on.[21]
152 153
The current recommendation from the South African National Institute of Communicable 154
Disease (NICD) is that suspected cases of pertussis in South Africa (SA) should have the 155
following tests performed for detection of B. pertussis:
156
• ideally: a nasopharyngeal swab or aspirate for PCR detection of B. pertussis.
157
• a nasopharyngeal swab or aspirate for culture of B. pertussis if PCR testing is not 158
feasible or available. [22]
159 160
The advantages and disadvantages of the available diagnostic methods are summarised in 161
Table 1 [22].
162 163
Table 1: Comparison of diagnostic methods for Bordetella pertussis
Method Advantage Disadvantage
Culture Highly specific – if positive it confirms the diagnosis. Can be done by most diagnostic microbiology laboratories pro- vided media and SOPs are available for processing.
Relatively cheap.
Poor sensitivity – highest in first two weeks (catarrhal phase) and is reduced following treatment. Higher sensitivity for infants than for adolescents and adults. Requires selective media and prolonged incubation (at least 7 days). Ideally culture medium should be inoculated at the bed- side.
Molecular techniques (PCR) Highly sensitive. Can detect B.
pertussis DNA even after treatment has commenced and remains positive late in the disease (≤3 weeks). Rapid results. At present the
recommended diagnostic test of choice if available.
Specificity can be a problem.
False positives do occur especially if only a single target PCR is used e.g. IS481. Requires molecular expertise and equipment. Relatively expensive.
Serology Relatively cheap and rapid test. Only useful if a standardized anti- PT IgG ELISA test is used (other antibodies lack sensitivity and specificity); even then, local cut- offs have not been determined.
Serology can NOT be used for diagnosis of pertussis in children or adults who have received acellular pertussis vaccine in the previous year, if not longer.
Not recommended alone for routine diagnosis.
Direct fluorescent antibody detection (DFA Rapid results. Poor sensitivity and specificity - many false negatives and false positives. Slides are difficult to interpret and prone to reader error. No longer recommended for routine diagnosis.
164
Accurate diagnosis of pertussis is important for the timely institution of optimal treatment 165
(a macrolide antibiotic) and infection control measures, especially in-hospital, and for 166
appropriate treatment of household contacts.
167 168
Recent evidence has shown that the finding of a pathogen or pathogens in respiratory 169
specimens does not always indicate a causal relationship. Organisms whose presence used 170
to be regarded as being of pathogenic significance are now found in otherwise healthy 171
subjects using new more sensitive molecular diagnostic tests. This has highlighted the 172
importance of the use of controls in studies looking at aetiological causes of respiratory 173
disease.[23]
174 175
Pertussis vaccine 176
The availability of an effective vaccine against Bordetella pertussis since the 1940’s has 177
substantially reduced the morbidity and mortality from this disease, preventing an 178
estimated 760 000 deaths annually. In many countries the original whole cell vaccine (wP) 179
has since been replaced by various formulations of the acellular vaccine (aP). However, 180
despite adequate vaccine coverage in many parts of the world, pertussis continues to 181
contribute a substantial burden of disease in un-immunised infants and increasingly 182
recognised infection and/or disease in adolescents and adults.[24] In the last decade there 183
appears to have been a substantial increase in pertussis cases amongst immunised 184
populations. The reasons for this are not fully elucidated but are in part due to improved 185
case detection and laboratory diagnostic procedures.[25] Recent evidence indicate that due 186
to the immune responses that aP vaccines induce that involve largely Th2 responses, they 187
may be less effective than wP vaccines that induce Th1 and Th17 responses. In addition, 188
the duration of protective immunity induced by aP vaccines is shorter than that induced by 189
wP.[26] South African infants were routinely immunised with the whole cell vaccine at 6, 190
10 and 14 weeks and boosted at 18 months of age as part of the National Expanded 191
Program on Immunisation (EPI) until April 2009 when this was changed to aP.[21]
192 193
In the Western Cape Province of South Africa, vaccine coverage in 2005 was found to be 194
80%, 77% and 48% for vaccines due by 14 weeks, 9 months and 18 months respectively.
195
Thus, a substantial number of children did not receive their early vaccines, while a large 196
proportion of children did not receive full courses of Diphtheria, Pertussis, Tetanus (DPT) 197
and measles vaccines. Children in the Boland region were significantly less likely to have 198
received vaccines due by both 14 weeks and 9 months compared to those in the Cape 199
Town Metro region.[27] Another study found vaccine coverage rates of 100%, 99% and 200
94% at 6, 10 and 14 weeks respectively in the Paarl area of the Western Cape between 201
2006 and 2008.[28] In another study, vaccine coverage had declined to 53% by the time of 202
the pertussis vaccine booster dose at 18 months.[29] There are not available reliable and 203
recent data – a survey is currently underway to collect this data.
204 205
The relative effectiveness of the vaccine in HIV-infected and HIV-exposed but uninfected 206
infants and children compared to HIV unexposed children is uncertain. In one 207
Cameroonian study, levels of antibodies against pertussis fimbrial antigens were 208
substantially lower in HIV-infected than in HIV-exposed but uninfected children and there 209
was a high risk of low antibody levels in response to the DTwP vaccine in those HIV- 210
infected children with severe immunodeficiency (CD4 T-cell level, <25%).[30, 31] The 211
concentrations of antibodies induced by the DTwP vaccine were lower in HIV-infected 212
children than in uninfected children. Likewise the quality and duration of immunity to 213
pertussis in HIV infected children once they are started on HAART is uncertain.[32] In a 214
cohort study conducted in Khayelitsha, Western Cape Province, South Africa that 215
included a review of antibodies against pertussis, the authors concluded that “Among 216
South African infants, antenatal HIV exposure was associated with lower specific 217
antibody responses in exposed uninfected infants compared with unexposed infants at 218
birth, but with robust responses following routine vaccination.”[33]
219 220
Surveillance Challenges 221
Potential obstacles to surveillance include a lack of standardised clinical case definitions, 222
making inter-country comparisons difficult, a lack of accurate diagnostic facilities for 223
confirmation of Bordetella pertussis in many developing countries (only two public health 224
laboratories currently offer the PCR diagnostic in South Africa), inadequate recognition 225
and reporting of cases by health care workers, particularly in adults and adolescents, and 226
the fact that passive notification systems significantly underestimate disease burden.
227 228
Rationale for the study 229
230
Hypothesis 231
We hypothesised that a substantial number of cases of severe childhood acute respiratory 232
infection in a South African hospital were due to Bordetella pertussis and Bordetella 233
parapertussis infection.
234 235
The aims:
236
1. To determine the burden of pertussis in infants and children with severe LRTI 237
2. To determine factors that are associated with increased risk of pertussis in children 238
with severe LRTI.
239
3. To determine the prevalence and type of respiratory co-infection in children 240
infected with confirmed pertussis 241
4. To develop a reliable clinical case definition of pertussis.
242
5. To conduct a systematic review of the epidemiological patterns of confirmed 243
pertussis in low- and middle-income countries since the inception of EPI in 1974.
244 245
A brief description of the cohort of participants used to answer the aims of the thesis 246
247
From September 2012 the study recruited children admitted for a lower respiratory tract 248
infection to the acute admission ward of the Red Cross War Memorial Children’s Hospital 249
in Cape Town, South Africa. The children were sequentially enrolled to a maximum of 250
four children per day over a full one-year period. Inclusion criteria were WHO-defined 251
age-specific tachypnoea or lower chest indrawing, apnoea. The children were less than 13 252
years (the age limit for children coming to the hospital). Children could only be enrolled if 253
parents were willing to sign informed consent.
254 255
We excluded participants if they had a previous admission to a health care facility in the 256
preceding two weeks. The reason for this was to minimise health care-associated infection 257
as study wanted primarily to assess community acquired pertussis.
258 259
A detailed history and clinical examination were done, especially noting the presence of 260
cough, apnoea, duration of symptoms and use of antibiotics prior to admission. History of 261
HIV exposure, infection and where relevant, antiretroviral treatment (ART) were 262
recorded. Information on immunisation was abstracted from the Road to Health Card 263
(RTHC), and the date and type of each vaccine recorded. The RTHC is a standardized 264
national record for each child.
265 266
HIV testing was done as appropriate for the age of the child and the children’s status 267
classified accordingly. Specific descriptions of both testing and classification appear in 268
each chapter where this is relevant.
269 270
Two nasopharyngeal (NP) swabs followed by an induced sputum (IS) specimen were 271
collected from each child and sent to the laboratory for both culture and PCR testing for 272
pertussis. In addition, a multiplex PCR for other respiratory pathogens was also 273
performed. As with HIV, descriptions of specific tests used, and their interpretations 274
appear in each relevant chapter.
275
276
As we also needed to assess the risk of pertussis posed by a close family member carrying 277
B. pertussis in their nasopharynx, the caregiver bringing the child was also enrolled for the 278
study. As with the child, the caregiver’s previous medical history (including HIV related 279
data) and history of recent symptoms were taken. An NP sample was taken from the 280
caregiver to be likewise tested for pertussis.
281 282
The study enrolled 460 child-caregiver pairs into the study. The data collected from these 283
enrolled participants were used to answer Aims 1 to 4 of the thesis as stated above. Brief 284
composition of the recruited participants is shown in Table 2.
285 286
Table 2: Baseline characteristics of enrolled participants (N=460)
Children Frequency n (%)
Age
< 2 months old 41 (8.9)
≥ 2 months old 419 (91.1)
Median (interquartile range) 7.8 (3.6-17.8) months Range 3.9 weeks - 12.7 years Gender
Female 202 (43.9) Male 258 (56.1) Number of samples
Nasopharyngeal Swabs 460 (100.0) Induced sputa 454 (98.7)
Caregivers
Age
Median (Interquartile range) 28 (24 - 33) years Range 15 - 52 years Relationship to child
Mother 450 (97.8) Father 2 (0.4) Grandmother 5 (1.1) Other 3 (0.7)
287 288 289
Outline of the thesis 290
291
General statement on the structure of the thesis 292
With the exception of Chapter 1, the introduction chapter, and Chapter 7, the conclusion 293
chapter, all the chapters in the thesis take the form of manuscripts that are either 294
published (in the case of Chapter 3) or undergoing peer review in various journals. Each 295
chapter contains its own literature review, methods and discussion sections, each relevant 296
to the specific aim of the thesis addressed by that chapter. As a result, the thesis does not 297
contain standalone literature review, methods, or discussion chapters. The thesis does 298
however contain a systematic review and metanalysis as detailed below. In addition, the 299
thesis has a short conclusion chapter, highlighting the findings of the thesis.
300 301
Each aim is dealt with separately in its own chapter with only the data necessary to 302
answer the aim specified for each chapter utilised as required in each instance. As a 303
result of this approach, numerators and denominators as well as summarised data, are 304
not always the same across all chapters, even when involving the same variables. This is 305
not an error. As an example, Chapter 3 which answers Aim 1 of the thesis includes both 306
Bordetella pertussis and Bordetella parapertussis confirmed cases in its numerator and 307
all 460 children in its denominator while Chapter 4, which is concerned with the risk of 308
pertussis due to Bordetella pertussis, and thus uses only confirmed Bordetella pertussis 309
its numerator to allow for comparison with other studies. Similarly, in Chapter 6, the 310
data of the few children above 9 years of age are excluded as they fell outside the ranges 311
of ages being considered for the diagnostic criteria considered in the analyses.”
312 313 314 315
Chapter 1 316
The background, rationale and outline of the thesis is presented. In the background, 317
the burden of pertussis is briefly described in the context of its changing epidemiology, 318
diagnostic and notification challenges, as well as vaccine coverage. South African 319
specific data is highlighted, indicting paucity thereof. In addition, a brief summary of 320
the methods and the participants is given. The chapter also includes an outline of the rest 321
of the chapters in the thesis. As each chapter contains its own literature review, the 322
chapter does not contain extensive literature review, but only what is essential to 323
establish grounds for this research.
324 325
Chapter 2 326
The chapter contains a formal systematic review on the burden of pertussis in LMICs.
327
The prevalence of pertussis is described stratified by geographic location, diagnostic 328
method, age categories as well as the period over which the cases were detected. The 329
chapter highlights the high case fatality rate in young infants as well as the increased 330
risk of pertussis burden posed by HIV infection and in utero exposure to HIV. The 331
systematic review describes both laboratory-confirmed Bordetella pertussis and 332
Bordetella parapertussis.
333 334
Chapter 3 335
In this chapter, the burden of pertussis in children admitted with acute lower respiratory 336
tract infections as defined by WHO is described. The chapter highlights the shorter 337
duration of symptoms at the time of diagnosis and describes an increased yield in 338
confirmed cases secondary to the use of a second specimen collected following induced 339
sputum. Also noted in this study is association of high risks of pertussis with HIV 340
exposure and infection.
341
Chapter 4 342
In this chapter, factors that flagged as potential risk factors in the previous 343
descriptive chapter are taken up and analysed further in more detail. A major 344
finding reported in this chapter is the high risk of confirmed pertussis in children whose 345
mothers have Bordetella pertussis isolated from a nasopharyngeal specimen. In 346
addition, the study analyses further the association with both HIV infection as well as 347
in-utero exposure to HIV uninfected children noted in the previous chapter by 348
quantifying the level of risk and establishing independence of risk. Additionally, the 349
study confirms in this African study, the well-known increased risk to pertussis 350
associated with incomplete immunisation, early infancy and poor nutritional status.
351 352
Chapter 5 353
The manuscript deals with bacterial and viral co-infections that occur with 354
pertussis. Here we describe the frequency of specific viral and bacterial 355
organism that are found in the lower respiratory tract of children investigated for 356
pertussis. The analysis includes correlating confirmed pertussis with the overall 357
number of coexisting potential pathogens as well as assessing the association 358
between pertussis and specific organisms. The importance of associated 359
respiratory pathogens detected with Bordetella pertussis are analysed with 360
respect to severity of respiratory symptoms.
361 362
Chapter 6 363
This manuscript assesses the sensitivity and specificity of clinical features compared to 364
PCR as reference standard in the diagnosis of pertussis. The chapter shows the poor 365
diagnostic accuracy of clinical case definitions and the limitation of these in both 366
clinical use and surveillance of pertussis. The addition of lymphocytosis to clinical 367
definitions is shown to be of limited value in improving diagnostic accuracy.
368
Chapter 7 369
As all relevant discussions are contained in each manuscript chapter, this is a short chapter 370
that reflects on findings and conclusions of the thesis. The chapter discusses the significance 371
of the findings from the systematic review and the four chapters reporting results from the 372
primary study. The resurgence of pertussis, highlighting the high mortality in young infants 373
and identified risk factors for pertussis are reflect on. The discussion brings into focus the 374
need for more awareness and need for improved diagnosis of pertussis. Finally, the focus 375
falls on the need for improved immunisation programs to control pertussis, especially 376
targeting high risk groups in the population.
377 378
Appendices: The following three documents have been appended to the end of the thesis:
379
Appendix A: Informed consent form 380
Appendix B: Ethical approval HREC 371/2011 381
Appendix C: Case Report Form 382
383
Author contributions to included manuscripts 384
385
The contributions to the manuscripts have been endorsed by my co-supervisors, 386
Professors Heather Zar and Gregory Hussey. All six manuscripts (published or under 387
review) have been approved by the University of Cape Town (UCT) doctoral degrees 388
board and UCT Vice chancellor as being appropriate for inclusion in the thesis as per 389
UCT policy. Permissions to include these manuscripts have be sought from and granted by 390
each of the co-authors involved in each manuscript.
391 392
1. The burden of laboratory confirmed pertussis in low- and middle-income countries 393
since the inception of the Expanded Programme on Immunisation (EPI) in 1974: a 394
systematic review and metanalysis. Muloiwa R, Kagina BM, Engel ME, Hussey 395
[Under review BMC Medicine]
396 397
Building on the published study protocol, I implemented the literature search strategy, 398
extracted the data and analysed it. B. Kagina assisted with the quality assurance required 399
for a systematic review study as per the study design and protocol. I analysed the data and 400
drafted the first manuscript. M. Engel reviewed the statistical analysis plan and results. G.
401
Hussey supervised all the aspects of the design and in the editing of the manuscript. The 402
final manuscript was approved by all the authors. This manuscript addresses Aim 5 of the 403
thesis.
404 405
2. Incidence and Diagnosis of Pertussis in South African Children Hospitalised With 406
Lower Respiratory Tract, Muloiwa R, Dube FS, Nicol MP, Zar HJ, Hussey GD. The 407
Pediatric infectious disease journal 2016; 35(6): 611-6.
408 409
I did the epidemiology study design for the project, including the analysis plan. G.
410
Hussey sourced the funding for the study. I managed the field data collection supervised 411
by H. Zar. F. Dube did the laboratory analysis of the specimens under supervision of M.
412
Nicol. I did all the data analysis and wrote the first draft of the paper, integrating 413
contributions from the co-authors. H. Zar and G. Hussey co-supervised the writing. All 414
authors provided contributions to the published manuscript. This manuscript talks to 415
Aim 1 of the thesis.
416 417
3. Impact of HIV status and maternal carriage on risk of childhood Bordetella 418
pertussis disease. Muloiwa R, Dube FS, Nicol MP, Hussey GD, Zar HJ [Under review 419
Plos One]
420 421
I designed the analysis plan to answer the question addressed by this manuscript using 422
the laboratory data supplied by F. Dube under supervision of M. Nicol. I did all the 423
data analysis and wrote the first draft of the paper, integrating contributions from the 424
co-authors. H. Zar and G. Hussey supervised and reviewed the manuscript. All authors 425
provided contributions to the published manuscript. The manuscript answers to Aim 2 of 426
the thesis.
427 428
4. Co-detection of Bordetella pertussis and other respiratory organisms in children 429
hospitalised with lower respiratory tract infection. Muloiwa R, Dube FS, Nicol MP, 430
Hussey GD, Zar HJ [Under review Scientific Reports]
431 432
I designed the analysis plan to answer the question addressed by this manuscript using 433
the laboratory data analysed by F. Dube under supervised by M. Nicol. I did all the 434
data analysis and wrote the first draft of the paper, integrating contributions from the 435
co-authors. H. Zar and G. Hussey supervised and reviewed the manuscript. All authors 436
provided contributions to the published manuscript. This manuscript address Aim 3 of 437
the thesis.
438 439
5. Diagnostic limitations of clinical case definitions of pertussis in infants and children 440
with severe lower respiratory tract infection. Muloiwa R, Nicol MP, Hussey GD, Zar 441
HJ [Under review Plos One]
442 443
I designed the analysis plan to answer the question addressed by this manuscript. I did 444
all the data analysis and wrote the first draft of the paper, integrating contributions from 445
the co-authors. M Nicol reviewed the manuscript while the final supervision was done 446
by H. Zar and G. Hussey. All authors provided contributions to the published 447
manuscript. Aim 4 of the thesis is addressed by this manuscript.
448 449 450 451
References 452
453
1. Crowcroft NS, Stein C, Duclos P, Birmingham M. How best to estimate the global burden 454
of pertussis? Lancet Infect Dis. 2003;3(7):413-8. Epub 2003/07/03.
455
2. World Health Organization. Managing pertussis outbreaks during humanitarian 456
emergencies : WHO technical note, February 2008. Geneva: World Health Organization;
457
2008. 6 p. p.
458
3. Jardine A, Conaty SJ, Lowbridge C, Staff M, Vally H. Who gives pertussis to infants?
459
Source of infection for laboratory confirmed cases less than 12 months of age during an 460
epidemic, Sydney, 2009. Communicable diseases intelligence quarterly report.
461
2010;34(2):116-21. Epub 2010/08/04.
462
4. Wood N, McIntyre P. Pertussis: review of epidemiology, diagnosis, management and 463
prevention. Paediatr Respir Rev. 2008;9(3):201-11; quiz 11-2. Epub 2008/08/13. doi:
464
10.1016/j.prrv.2008.05.010.
465
5. Cherry JD. Pertussis: challenges today and for the future. PLoS pathogens.
466
2013;9(7):e1003418. Epub 2013/08/13. doi: 10.1371/journal.ppat.1003418.
467
6. Muloiwa R, Moodley M, Zar H, editors. Modifying the clinical case definition of pertussis 468
increases the sensitivity of diagnosis in children suspected of Bordetella pertussis 469
infection 16th International Congress on Infectious Diseases (ICID); 2014; Cape Town, 470
South Africa.
471
7. Guiso N, Hegerle N. Other Bordetellas, lessons for and from pertussis vaccines. Expert 472
review of vaccines. 2014;13(9):1125-33.
473 8. Mattoo S, Cherry JD. Molecular pathogenesis, epidemiology, and clinical manifestations 474
of respiratory infections due to Bordetella pertussis and other Bordetella subspecies.
475
Clinical microbiology reviews. 2005;18(2):326-82. Epub 2005/04/16. doi:
476
10.1128/cmr.18.2.326-382.2005.
477 9. World Health Organisation. WHO Immunological Basis for Immunization Series Module 478
4: Pertussis: Geneva: World Health Organization;; 2017. Available from:
479
https://apps.who.int/iris/handle/10665/259388.
480 10. Paddock CD, Sanden GN, Cherry JD, Gal AA, Langston C, Tatti KM, et al. Pathology 481
and pathogenesis of fatal Bordetella pertussis infection in infants. Clinical infectious 482
diseases : an official publication of the Infectious Diseases Society of America.
483
2008;47(3):328-38. Epub 2008/06/19. doi: 10.1086/589753.
484
11. World Health Organisation. WHO-recommended surveillance standard of pertussis [11 485
July 2019)]. Available from:
486
http://www.who.int/immunization/monitoring_surveillance/burden/vpd/surveillance_type/
487
passive/pertussis_standards/en/.
488
12. Centers for Disease Control. Pertussis / Whooping Cough (Bordetella pertussis) 2020 489
Case Definition 2020 [29 November 2020]. Available from:
490
https://wwwn.cdc.gov/nndss/conditions/pertussis/case-definition/2020/.
491
13. Wittenberg D. Standard Treatment Guidelines and Essential Drugs List For South Africa, 492
Paediatric Hospital Level. Second ed: National Department of Health; 2006. p. 188.
493
14. Hallbauer UM, Pieters M, Elliott E, Y G. Ongoing Bordetella infection since April 2008 494
in Bloemfontein, South Africa. 7th World Congress of the World Society for Pediatric 495
Infectious Diseases; Melbourne, Australia 2011.
496
15. Tatti KM, Sparks KN, Boney KO, Tondella ML. Novel multitarget real-time PCR assay 497
for rapid detection of Bordetella species in clinical specimens. J Clin Microbiol.
498
2011;49(12):4059-66. Epub 2011/09/24. doi: 10.1128/jcm.00601-11.
499
16. Njamkepo E, Bonacorsi S, Debruyne M, Gibaud SA, Guillot S, Guiso N. Significant 500
finding of Bordetella holmesii DNA in nasopharyngeal samples from French patients with 501
suspected pertussis. J Clin Microbiol. 2011;49(12):4347-8. Epub 2011/10/21. doi:
502
10.1128/jcm.01272-11.
503
17. Crowcroft NS, Pebody RG. Recent developments in pertussis. Lancet.
504
2006;367(9526):1926-36. Epub 2006/06/13. doi: S0140-6736(06)68848-X [pii]
505
10.1016/S0140-6736(06)68848-X [doi].
506
18. Shankland I, Corcoran C, Zar H, Diedericks R, Whitelaw A. Detection of Bordetella 507
